Connect with us

Artificial Intelligence

OpenAI Releases Codex Plugin That Runs Inside Anthropic’s Claude Code

mm

OpenAI published a Codex plugin on March 30 that installs directly inside Anthropic’s Claude Code, letting developers run code reviews and delegate tasks to Codex without leaving their existing workflow. The open-source plugin, released under an Apache 2.0 license, is the first official OpenAI integration designed to run inside a rival’s coding environment.

The plugin provides six slash commands. /codex:review runs a standard read-only Codex code review. /codex:adversarial-review adds a steerable challenge mode that questions implementation decisions, tradeoffs, and failure modes. /codex:rescue hands work off to Codex entirely, operating as a subagent that can investigate bugs, attempt fixes, or take a second pass at a problem. Three additional commands — /codex:status, /codex:result, and /codex:cancel — manage background jobs.

Installation requires a ChatGPT subscription (including the free tier) or an OpenAI API key, plus Node.js 18.18 or later. Codex usage through the plugin counts against existing Codex usage limits.

How the Plugin Works

Rather than bundling a separate runtime, the plugin delegates through the local Codex CLI and Codex app server already installed on a developer’s machine. It reuses the same authentication, configuration, environment variables, and MCP server setup that Codex uses directly. Project-level settings in .codex/config.toml and user-level settings in ~/.codex/config.toml both apply.

The plugin also includes an optional review gate feature. When enabled via /codex:setup --enable-review-gate, Codex automatically reviews Claude’s output before it finalizes. If the review finds issues, it blocks completion so Claude can address them first. OpenAI’s documentation warns this feature can create long-running loops and drain usage limits quickly.

The release comes days after OpenAI launched a broader plugin marketplace for Codex, which added integrations with Slack, Notion, Figma, Gmail, and Google Drive. That system lets developers package skills, app integrations, and MCP server configurations into installable bundles distributed across teams. The Codex plugin directory now lists more than 20 available plugins across the app, CLI, and VS Code extension.

A Shift in Competitive Dynamics

OpenAI building an official plugin for a competitor’s platform marks a notable shift. The company has been consolidating its developer tools into a unified desktop experience, but this release extends Codex’s reach into the workflow of developers who have chosen Anthropic’s tooling instead.

The move reflects a practical reality in the AI coding tool market: developers often use multiple assistants, and locking them into a single ecosystem has proven difficult. By making Codex available inside Claude Code, OpenAI gains visibility with developers who might not otherwise interact with its coding tools, while those developers get access to a second model’s perspective on their code.

The timing also coincides with Claude’s surging popularity among paying users. Claude Code has become a significant distribution channel, and OpenAI appears to be meeting those developers where they already work rather than asking them to switch.

The growing competition among AI coding tools has pushed companies toward interoperability rather than walled gardens. Whether other AI tool makers follow OpenAI’s lead and build plugins for rival platforms remains an open question – but the precedent is now set.

Alex McFarland is an AI journalist and writer exploring the latest developments in artificial intelligence. He has collaborated with numerous AI startups and publications worldwide.